[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Where to put (unix) DLMs?

In article <onitjc4h0k.fsf@cow.physics.wisc.edu>,
craigmnet@cow.physics.wisc.edu wrote something like this:

> Let's say that I am distributing a package with a DLM.  Of course the
> problem with DLMs is they can be both IDL version-dependent, and
> architecture dependent.  At least mine is. :-) Let's say for the
> moment can make a big library of binary DLM files for all the
> architectures and IDL versions I know about.
Ouch! For rmfit, we ended up distributing source code for the *nix users
and asked them to make the DLM themselves. The Mac binary is available at
special request; as for Windows...

> The question is, where should the DLMs?  The nicest place would be the
> IDL binary directory.  Of course that assumes that users have
> permission to put things there.
For Macs, this is the only choice I can see. Unix is a lot harder, for
reasons you've mentioned below. In our distribution, the DLM goes in a
subdirectory of the source tree and users have to set up their
IDL_DLM_PATH. I put all this in the installation notes and hope they will
only have to do this once.

> Another thought is to have a separate "site" DLM directory.  That
> requires the user to set their IDL_DLM_PATH environment variable
> before running IDL.  This works, unless users are in heterogeneous
> environment (ie, like my lab's), in which case they won't know *which*
> directory to set it to ahead of time.  Sadly, the IDL system variable
> !DLM_PATH cannot be set at runtime, so there is no way to discover it
> on the fly.

Too true, perhaps it's worth a note to RSI to put in a change request? 


            - Rob Preece 
              Dept. of Physics
              University of Alabama in Huntsville

This line no sig